swin-tiny-patch4-window7-224_alpha0.5_temp3.0_t3

This model is a fine-tuned version of c14kevincardenas/ClimBEiT-t3 on the c14kevincardenas/beta_caller_284_person_crop_seq_withlimb dataset. It achieves the following results on the evaluation set:

• Loss: 0.6361
• Accuracy: 0.8341

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.8113	1.0	164	1.4079	0.2570
0.7122	2.0	328	1.1494	0.4685
0.5814	3.0	492	0.9148	0.6681
0.4872	4.0	656	0.8063	0.7299
0.4316	5.0	820	0.7516	0.7516
0.3833	6.0	984	0.7281	0.7852
0.3552	7.0	1148	0.7019	0.8080
0.3451	8.0	1312	0.6698	0.8124
0.3342	9.0	1476	0.6744	0.8113
0.3062	10.0	1640	0.6680	0.8080
0.3047	11.0	1804	0.6803	0.8145
0.3064	12.0	1968	0.6599	0.8221
0.2918	13.0	2132	0.6601	0.8297
0.3017	14.0	2296	0.6374	0.8254
0.2998	15.0	2460	0.6361	0.8341
0.2955	16.0	2624	0.6393	0.8341
0.2946	17.0	2788	0.6496	0.8341
0.288	18.0	2952	0.6405	0.8308
0.2928	19.0	3116	0.6427	0.8330
0.2853	20.0	3280	0.6402	0.8330

Framework versions

• Transformers 4.45.2
• Pytorch 2.5.0+cu124
• Datasets 3.0.1
• Tokenizers 0.20.1